Learning graphs to model visual objects across different depictive styles

Q. Wu, H. Cai, P. Hall

Research output: Chapter in Book/Report/Conference proceedingChapter

  • 4 Citations

Abstract

Visual object classification and detection are major problems in contemporary computer vision. State-of-art algorithms allow thousands of visual objects to be learned and recognized, under a wide range of variations including lighting changes, occlusion, point of view and different object instances. Only a small fraction of the literature addresses the problem of variation in depictive styles (photographs, drawings, paintings etc.). This is a challenging gap but the ability to process images of all depictive styles and not just photographs has potential value across many applications. In this paper we model visual classes using a graph with multiple labels on each node; weights on arcs and nodes indicate relative importance (salience) to the object description. Visual class models can be learned from examples from a database that contains photographs, drawings, paintings etc. Experiments show that our representation is able to improve upon Deformable Part Models for detection and Bag of Words models for classification.
LanguageEnglish
Title of host publicationComputer Vision – ECCV 2014
Subtitle of host publication13th European Conference, Zurich, Switzerland, September 6-12, 2014, Proceedings, Part VII
EditorsDavid Fleet , Tomas Pajdla , Bernt Schiele , Tinne Tuytelaars
Place of PublicationCham, Switzerland
PublisherSpringer
Pages313-328
Number of pages16
ISBN (Print)9783319105833
DOIs
StatusPublished - 22 Sep 2014
Event13th European Conference on Computer Vision, ECCV 2014; Zurich - Zurich , Switzerland
Duration: 6 Sep 201412 Sep 2014

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume8695

Conference

Conference13th European Conference on Computer Vision, ECCV 2014; Zurich
CountrySwitzerland
CityZurich
Period6/09/1412/09/14

Fingerprint

Painting
Computer vision
Labels
Lighting
Experiments

Cite this

Wu, Q., Cai, H., & Hall, P. (2014). Learning graphs to model visual objects across different depictive styles. In D. Fleet , T. Pajdla , B. Schiele , & T. Tuytelaars (Eds.), Computer Vision – ECCV 2014: 13th European Conference, Zurich, Switzerland, September 6-12, 2014, Proceedings, Part VII (pp. 313-328). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 8695). Cham, Switzerland: Springer. DOI: 10.1007/978-3-319-10584-0_21

Learning graphs to model visual objects across different depictive styles. / Wu, Q.; Cai, H.; Hall, P.

Computer Vision – ECCV 2014: 13th European Conference, Zurich, Switzerland, September 6-12, 2014, Proceedings, Part VII. ed. / David Fleet ; Tomas Pajdla ; Bernt Schiele ; Tinne Tuytelaars . Cham, Switzerland : Springer, 2014. p. 313-328 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 8695).

Research output: Chapter in Book/Report/Conference proceedingChapter

Wu, Q, Cai, H & Hall, P 2014, Learning graphs to model visual objects across different depictive styles. in D Fleet , T Pajdla , B Schiele & T Tuytelaars (eds), Computer Vision – ECCV 2014: 13th European Conference, Zurich, Switzerland, September 6-12, 2014, Proceedings, Part VII. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 8695, Springer, Cham, Switzerland, pp. 313-328, 13th European Conference on Computer Vision, ECCV 2014; Zurich, Zurich , Switzerland, 6/09/14. DOI: 10.1007/978-3-319-10584-0_21
Wu Q, Cai H, Hall P. Learning graphs to model visual objects across different depictive styles. In Fleet D, Pajdla T, Schiele B, Tuytelaars T, editors, Computer Vision – ECCV 2014: 13th European Conference, Zurich, Switzerland, September 6-12, 2014, Proceedings, Part VII. Cham, Switzerland: Springer. 2014. p. 313-328. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). Available from, DOI: 10.1007/978-3-319-10584-0_21
Wu, Q. ; Cai, H. ; Hall, P./ Learning graphs to model visual objects across different depictive styles. Computer Vision – ECCV 2014: 13th European Conference, Zurich, Switzerland, September 6-12, 2014, Proceedings, Part VII. editor / David Fleet ; Tomas Pajdla ; Bernt Schiele ; Tinne Tuytelaars . Cham, Switzerland : Springer, 2014. pp. 313-328 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).
@inbook{319380ef59184b06a41124d61d9b9012,
title = "Learning graphs to model visual objects across different depictive styles",
abstract = "Visual object classification and detection are major problems in contemporary computer vision. State-of-art algorithms allow thousands of visual objects to be learned and recognized, under a wide range of variations including lighting changes, occlusion, point of view and different object instances. Only a small fraction of the literature addresses the problem of variation in depictive styles (photographs, drawings, paintings etc.). This is a challenging gap but the ability to process images of all depictive styles and not just photographs has potential value across many applications. In this paper we model visual classes using a graph with multiple labels on each node; weights on arcs and nodes indicate relative importance (salience) to the object description. Visual class models can be learned from examples from a database that contains photographs, drawings, paintings etc. Experiments show that our representation is able to improve upon Deformable Part Models for detection and Bag of Words models for classification.",
author = "Q. Wu and H. Cai and P. Hall",
year = "2014",
month = "9",
day = "22",
doi = "10.1007/978-3-319-10584-0_21",
language = "English",
isbn = "9783319105833",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
publisher = "Springer",
pages = "313--328",
editor = "{Fleet }, {David } and {Pajdla }, {Tomas } and {Schiele }, {Bernt } and {Tuytelaars }, {Tinne }",
booktitle = "Computer Vision – ECCV 2014",

}

TY - CHAP

T1 - Learning graphs to model visual objects across different depictive styles

AU - Wu,Q.

AU - Cai,H.

AU - Hall,P.

PY - 2014/9/22

Y1 - 2014/9/22

N2 - Visual object classification and detection are major problems in contemporary computer vision. State-of-art algorithms allow thousands of visual objects to be learned and recognized, under a wide range of variations including lighting changes, occlusion, point of view and different object instances. Only a small fraction of the literature addresses the problem of variation in depictive styles (photographs, drawings, paintings etc.). This is a challenging gap but the ability to process images of all depictive styles and not just photographs has potential value across many applications. In this paper we model visual classes using a graph with multiple labels on each node; weights on arcs and nodes indicate relative importance (salience) to the object description. Visual class models can be learned from examples from a database that contains photographs, drawings, paintings etc. Experiments show that our representation is able to improve upon Deformable Part Models for detection and Bag of Words models for classification.

AB - Visual object classification and detection are major problems in contemporary computer vision. State-of-art algorithms allow thousands of visual objects to be learned and recognized, under a wide range of variations including lighting changes, occlusion, point of view and different object instances. Only a small fraction of the literature addresses the problem of variation in depictive styles (photographs, drawings, paintings etc.). This is a challenging gap but the ability to process images of all depictive styles and not just photographs has potential value across many applications. In this paper we model visual classes using a graph with multiple labels on each node; weights on arcs and nodes indicate relative importance (salience) to the object description. Visual class models can be learned from examples from a database that contains photographs, drawings, paintings etc. Experiments show that our representation is able to improve upon Deformable Part Models for detection and Bag of Words models for classification.

UR - http://www.scopus.com/inward/record.url?scp=84906352971&partnerID=8YFLogxK

UR - http://dx.doi.org/10.1007/978-3-319-10584-0_21

UR - http://eccv2014.org/

U2 - 10.1007/978-3-319-10584-0_21

DO - 10.1007/978-3-319-10584-0_21

M3 - Chapter

SN - 9783319105833

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 313

EP - 328

BT - Computer Vision – ECCV 2014

PB - Springer

CY - Cham, Switzerland

ER -